/* ReCoder.h Recodes different Russian tables: ASCII(a,b), MIC, KOI-8, KOI-7, ISO-8859-5 and different new line symbols: CR+LF <-> CR. NOTE: If you want to translate from SRC-code to DST-code, you may use tr///; Note, that tables are swapped. */ #include #include #include #ifndef BYTE_DEF typedef unsigned char BYTE; #define BYTE_DEF #endif #define QUESTION (BYTE)'?' #define CONTROL_Z (BYTE)26 #define ASCII_128 (BYTE)'' #define ASCII_32 (BYTE)' ' #define ASCII_64 (BYTE)'@' #if defined (_MSDOS_) || defined (_MS_DOS_) || defined (__MSDOS__) #ifndef __MSDOS__ #define __MSDOS__ #endif #else #ifndef __UNIX__ #define __UNIX__ #endif #endif #ifdef __UNIX__ #define STRCMP strcmp #define stdhelp stderr #else #define STRCMP stricmp #define stdhelp stdout #endif #ifndef stricmp #define stricmp strcmp #endif #define ERROR(s) fprintf(stderr,"%s: %s\n",argv[0],s),exit(1) #define ERROR2(s,t) fprintf(stderr,"%s: %s %s\n",argv[0],s,t),exit(1) const char HELPTEXT[] = "\n" "Written by Alexey Roytman.\n" "Version 2.0\n" "Last modified: Apr 1 1998\n" "\n" "Recodes different Russian code tables and\n" "UNIX/DOS new line codes ('\\n'<->'\\r'+'\\n')\n" " and \"end-of-file\" symbols (''<->'^Z')\n\n" "Usage:\n" "\tReCoder [options] [infile [outfile]]\n" "\t\twhere 'infile' and 'outfile' are filenames or devicenames,\n" "\t\tor '--' for standart input-output device\n\n" "or\n" "\tReCoder [options] [ < instream] [ > outstream]\n" "\t\twhere 'instream' and 'outstream' are filenames or devicenames.\n\n" "'options' -- either sequence or one of the following types:\n" "\n" "\t-help\t-- this help\n" "\n" "\t-2\n" "\t\twhere and are both from the set:\n" "\t\ta\t-- alternative\n" "\t\tk\t-- KOI-8\n" "\t\tb\t-- Besta\n" "\t\t7\t-- KOI-7\n" "\t\tl\t-- KOI-7 uppercase-downcase inversed (latin transliterated)\n" "\t\tw\t-- Windows\n" "\t\ti\t-- ISO-8859-5\n" "\t\tg\t-- MIC Bulgarian\n" "\t\to\t-- main\n" "\t\tm\t-- Macintosh\n" "\n" "\t\t(for example: totally 8x8=64 possibilities of such pairs\n" "\t\t [such as: a2k, k2a, a2w, w2a, k2w, w2k, ...]\n" "\t\t are acceptable, including the tautologies\n" "\t\t [such as: a2a, k2k, w2w, ...]\n" "\t\t that produce \"letter-to-itself\" translation\n" "\n" "\t-u2d\t-- UNIX -> DOS\n" "\t-d2u\t-- DOS -> UNIX (in DOS works only for files, not for streams).\n" "\n" "\n" "Usable [i.e. I use them frequently] examples of parameters:\n" "\n" "\t-a2k -d2u \\\n" "\t -- translate de-facto Russian standards: from DOS to *NIX\n" "\t-d2u -a2k /\n" "\t-k2a -u2d \\\n" "\t -- translate de-facto Russian standards: from *NIX to DOS\n" "\t-u2d -a2k /\n" "\t-a2w -- translate alternative encoding to Windows one...\n" "\t-w2a -- ... and vice versa\n" "\n" "\t-k2l -- translate KOI-8 to pseudo-transliterated Russian\n" "\n" "\n" "Additional information:\n" "table sample: \"[ABV.\\EJuJa.abv.\\ejuja]\"\n" "\n" "Alternative -- Brjabrin's \"al'ternativnaya kodirovka\";\n" " Microsoft DOS Code Page 866 compatible\n" "[\"...\"]\n" "\n" "ISO-8859-5 -- Standard 8859-5 by International Standard Organization\n" "[\"...\"]\n" "\n" "Main -- Brjabrin's \"osnovnaya kodirovka\";\n" " GOST main standard\n" "[\"...\"]\n" "\n" "MIC -- \"standard\" before GOST;\n" " \"Bulgarian\" code table\n" "[\"...\"]\n" "\n" "KOI-8 -- KOI-8 according to RFC-1489 (1993) standard\n" " (by A.Chernov, RELCOM Development Team)\n" "[\"...\"]\n" "\n" "KOI-7 -- 7-bit\n" "[\"abw.|`q.ABW.\\@Q\"]\n" "\n" "KOI-7 inv -- 7-bit uppercase-downcase inversed.\n" " Useful for \"Russian->Transliterated\" translation.\n" "[\"ABW.\\@Q.abw.|`q\"]\n" "\n" "Besta -- as KOI-8, but Russian upper- and lower-case\n" " letters interchanged\n" "[\"...\"]\n" "\n" "Windows -- Microsoft Windows Code Page 1251\n" "[\"...\"]\n" "\n" "Macintosh -- Apple Macintosh\n" "[\"...\"]\n" "\n" "\n" "-----------------------------------------------------------------------\n" ; /* "GOST alternative" and "GOST alternative modified" are the same main part but different at the end (last 16 symbols are irrelevant). (by V.M.Brjabrin and V.E.Figurnov) */ const BYTE Tbl_GOST_alt[]= "" "" "" "" "" "" "" ""; /* "GOST main" is rather different from "GOST alternative". (by V.M.Brjabrin and V.E.Figurnov) */ const BYTE Tbl_GOST_main[]= /* "ѵս" "ɻͺ˹ΰ" "ڿij´" "" "" "" "" ""; */ "" "" "" "" "" "" "" ""; /* "MIC" or "bulgar" coding (the basic table before GOST). Caution: the symbols with codes 224239 are irrelevant. (by V.E.Figurnov) */ const BYTE Tbl_MIC[]= /* "" "" "" "" "Źο" "" " " ""; */ "" "" "" "??????" "???" "?????????" "" ""; /* "KOI-8" coding table. (according to RFC 1489 by A.Chernov) */ const BYTE Tbl_KOI8[]= /* "ijڿô" "" "ͺɸȾ" "̵" "椥娩" "㦢" "" ""; */ "" "" "" "" "" "" "" ""; /* "Besta" -- "KOI-8 inversed" coding table. */ const BYTE Tbl_BESTA[]= /* "ijڿô" "" "ͺɸȾ" "̵" "" "" "椥娩" "㦢"; */ "" "" "" "" "" "" "" ""; /* MS-Windows Code Page 1251. */ const BYTE Tbl_CP1251[]= /* " " " " " " " " "" "" "" ""; */ "" "" "" " " " " " " "" " "; /* ISO-8859-5. */ const BYTE Tbl_ISO8859_5[]= /* " " " " " " "" "" "" "" " "; */ "" "" "" " " " " " " "" " "; /* KOI-7. */ const BYTE Tbl_KOI7[]= "abwgdevzijklmnop" "rstufhc~{}\"yx|`q" "ABWGDEVZIJKLMNOP" " " " " " " "RSTUFHC^[]_YX\\@Q" "$# "; /* KOI-7 inversed. */ const BYTE Tbl_KOI7_inv[]= "ABWGDEVZIJKLMNOP" "RSTUFHC^[]_YX\\@Q" "abwgdevzijklmnop" " " " " " " "rstufhc~{}\"yx|`q" "#$ "; /* Macintosh. */ const BYTE Tbl_Macintosh[]= "" "" "" "" "" "" "" ""; /**/